Source Count: 21 | Weighted Score: 35 | Source Confidence: [4/5] | Primary Tier: 1 | Last Updated: March 11, 2026
Keywords: knowledge representation, ontology, semantic web, knowledge graph, RDF, OWL, description logic, frames, reasoning, artificial intelligence
Category Tags: information-computation, artificial-intelligence, semantic-web, data-science, logic
Cross-References: ZD_5_10 — Information Retrieval · ZD_5_07 — Search Algorithms · ZD_1_02 — Mathematics Information

QUICK SUMMARY

Knowledge representation (KR) is the field of artificial intelligence concerned with how to formally encode information about the world — facts, relationships, concepts, rules, and constraints — in formats that computer systems can use for reasoning, inference, and problem solving. It is one of the oldest and most fundamental problems in AI: how do you represent "the sky is blue," "all dogs are mammals," "aspirin treats headaches," or "Paris is the capital of France" in a way that a machine can store, query, combine with other knowledge, and use to derive new conclusions? The field has produced multiple representational paradigms: (1) Logic-based representations — propositional logic, first-order predicate logic, description logics — expressing knowledge as formal statements amenable to automated reasoning; (2) Frames (Minsky, 1975) — structured representations of stereotypical situations, with slots for attributes and default values, influential in object-oriented programming; (3) Semantic networks — graph-based representations where nodes are concepts and edges are relationships; (4) Ontologies — formal, explicit specifications of shared conceptualizations (Gruber, 1993) — hierarchical structures defining types, properties, and relationships within a domain; the Semantic Web vision (Berners-Lee, Hendler, and Lassila, 2001) proposed building a machine-readable layer on top of the existing web using RDF (Resource Description Framework — representing knowledge as subject-predicate-object triples), OWL (Web Ontology Language — for defining classes, properties, and axioms enabling automated reasoning), and SPARQL (query language for RDF); while the full Semantic Web vision did not materialize as envisioned, its technologies underpin knowledge graphs — large-scale structured representations of entities and their relationships used by Google (Knowledge Graph, 2012 — "things not strings"), Facebook, Amazon, Microsoft (Satori), and Wikidata; (5) Knowledge graphs have become a critical infrastructure component — Google's Knowledge Graph contains billions of facts about hundreds of millions of entities, powering search results, featured snippets, and Google Assistant; DBpedia and Wikidata provide open, collaboratively maintained knowledge graphs derived from Wikipedia. Modern KR increasingly integrates symbolic representations (ontologies, knowledge graphs) with neural approaches (knowledge graph embeddings — TransE, DistMult — representing entities and relations as vectors in continuous space), attempting to combine the precision and explainability of symbolic AI with the flexibility and scalability of neural AI — a central theme in the "neuro-symbolic AI" research agenda.

1. VERIFIED CLAIMS (Tier 1 — Peer-Reviewed / Established)

1.1 Classical Knowledge Representation

• Frames (Minsky, 1975): structured representations of stereotypical objects or situations with slots for attributes, default values, and inheritance relationships; directly influenced object-oriented programming and database schema design
• Semantic networks (Quillian, 1968): graph structures representing conceptual knowledge — nodes for concepts, labeled directed edges for relationships (IS-A, PART-OF, HAS-PROPERTY); foundational for later knowledge graph approaches
• Description logics: a family of formal knowledge representation languages that provide a logical basis for ontologies — well-defined semantics, decidable reasoning (subsumption, satisfiability, classification); the theoretical foundation of OWL (OWL DL corresponds to the description logic SROIQ)

1.2 Semantic Web Technologies

• RDF (Resource Description Framework): W3C standard for representing information as subject-predicate-object triples (e.g., "Paris" — "capitalOf" — "France"); designed for machine-readable linked data on the web; enables federation — data from different sources can be combined when using shared URIs
• OWL (Web Ontology Language): W3C standard for defining ontologies with rich class hierarchies, property constraints, and axioms enabling automated reasoning (classification, consistency checking); widely used in biomedical informatics (SNOMED CT, Gene Ontology), cultural heritage (CIDOC-CRM), and industry

1.3 Knowledge Graphs

• Google Knowledge Graph (Singhal, 2012): "things not strings" — structured database of entities (people, places, organizations, events, concepts) and their relationships; powers knowledge panels in Google Search, featured snippets, voice assistant answers; contains billions of facts about hundreds of millions of entities
• Wikidata (2012–): free, collaborative, multilingual knowledge graph operated by the Wikimedia Foundation; contains ~100 million items with structured data; serves as a centralized knowledge base for Wikipedia and is widely used in research and applications

2. CREDIBLE CLAIMS (Tier 2 — Academic / Debated but Supported)

2.1 Knowledge Graph Embeddings

• TransE (Bordes et al., 2013): representing entities and relations as vectors in continuous space such that for a true triple (head, relation, tail), head + relation ≈ tail; enables link prediction, knowledge base completion, and integration with neural networks; followed by TransR, DistMult, ComplEx, RotatE — each handling different relation patterns (symmetry, composition, one-to-many)
• Neuro-symbolic integration: combining knowledge graphs with neural networks — knowledge-enhanced language models (injecting structured knowledge into transformers), graph neural networks operating on knowledge graphs, using LLMs to extract and populate knowledge graphs from text; actively researched, results promising but no dominant paradigm

2.2 Ontology Engineering Challenges

• Ontology alignment: matching concepts across different ontologies that describe the same domain differently — automated matching is difficult due to differences in granularity, terminology, and modeling choices; critical for data integration across organizations
• Knowledge acquisition bottleneck: building and maintaining large ontologies requires significant domain expertise and effort; automated ontology learning from text and data reduces but does not eliminate this challenge

3. SPECULATIVE CLAIMS (Tier 3 — Possible but Unverified)

3.1 LLMs as Knowledge Bases

• Implicit knowledge in language models: large language models contain vast amounts of world knowledge encoded implicitly in their parameters — researchers argue LLMs could replace explicit knowledge graphs for many applications; however, LLMs lack the precision, verifiability, provenance tracking, and updateability of structured knowledge graphs; the relationship is more likely complementary than substitutive

4. DUBIOUS CLAIMS (Tier 4 — No Credible Source / Contradicted by Evidence)

4.1 The Semantic Web Is Dead

• [MISLEADING] While the grand vision of a fully machine-readable web has not been realized as originally envisioned by Berners-Lee, Semantic Web technologies (RDF, OWL, SPARQL) are widely used in practice — schema.org is embedded in millions of websites, knowledge graphs power major search engines and virtual assistants, linked data is standard in biomedicine and government data; the technologies succeeded even if the vision was partially realized

COUNTER-ARGUMENTS

• Neuro-symbolic integration debate: Whether explicit symbolic knowledge representation (ontologies, knowledge graphs) remains necessary in the age of LLMs is contested. Yann LeCun and neural network proponents have suggested that learned representations can subsume symbolic knowledge, while Gary Marcus and Yoshua Bengio (in his later work) have argued that hybrid neuro-symbolic systems are needed for robust reasoning, compositionality, and out-of-distribution generalization
• Ontology engineering brittleness: CYC (Lenat) — the decades-long project to encode common-sense knowledge in formal logic — is regarded by many as a cautionary tale about the difficulty of comprehensive knowledge engineering. Critics argue that the combinatorial explosion of edge cases and context-dependence makes complete formal ontologies unachievable, while defenders note that domain-specific ontologies (Gene Ontology, SNOMED CT) have proven practically valuable
• Knowledge graph completeness: Real-world knowledge graphs (Wikidata, DBpedia) are known to be highly incomplete — Krompaß, Baier, and Tresp (2015) estimated that major KGs capture only a small fraction of potentially valid triples, and completion methods introduce statistical biases

IMAGES

No images assigned yet.

BIBLIOGRAPHY

1. Minsky, Marvin | 1975 | "A Framework for Representing Knowledge" | The Psychology of Computer Vision | ∅ | ∅ | In , edited by Patrick Henry Winston, 211 277 | ∅ | | ∅ | ∅ | New York: McGraw-Hill
2. Berners-Lee, Tim, James Hendler; Ora Lassila | 2001 | "The Semantic Web" | Scientific American | ∅ | 284.5::34–43 | ∅ | ∅ | doi:10.1038/scientificamerican0501-34 | ∅ | ∅ | ∅
3. Gruber, Thomas R | 1993 | "A Translation Approach to Portable Ontology Specifications" | Knowledge Acquisition | ∅ | 5.2::199–220 | ∅ | ∅ | doi:10.1006/knac.1993.1008 | ∅ | ∅ | ∅
4. Baader, Franz, et al., eds. . | 2007 | ∅ | The Description Logic Handbook | ∅ | ∅ | Cambridge: Cambridge University Press | 2nd | doi:10.1007/s11023-004-4929-2 | ∅ | ∅ | ∅
5. Hogan, Aidan, et al | 2022 | "Knowledge Graphs" | ACM Computing Surveys | ∅ | 54.4:: | Article 71 | ∅ | ∅ | ∅ | ∅ | ∅
6. Bordes, Antoine, et al. : 2787 2795 | 2013 | "Translating Embeddings for Modeling Multi-Relational Data" | NIPS | ∅ | ∅ | ∅ | ∅ | ∅ | ∅ | ∅ | ∅
7. Singhal, Amit | 2012 | "Introducing the Knowledge Graph" | ∅ | ∅ | ∅ | Google Blog, May 16 | ∅ | ∅ | ∅ | ∅ | ∅
8. Brachman, Ronald J.; Hector J | 2004 | ∅ | Knowledge Representation and Reasoning | ∅ | ∅ | Levesque | ∅ | doi:10.1145/15923.1058024 | ∅ | ∅ | San Francisco: Morgan Kaufmann
9. Brachman, Ronald J.; Hector J | 2004 | ∅ | Knowledge Representation and Reasoning | ∅ | ∅ | Levesque | ∅ | doi:10.1145/15923.1058024 | ∅ | ∅ | San Francisco: Morgan Kaufmann
10. Berners-Lee, Tim, James Hendler; Ora Lassila | 2001 | "The Semantic Web" | Scientific American | ∅ | 284.5::34–43 | ∅ | ∅ | ∅ | ∅ | ∅ | ∅
11. Hogan, Aidan, et al | 2021 | "Knowledge Graphs" | ACM Computing Surveys | ∅ | 54.4::1–37 | ∅ | ∅ | ∅ | ∅ | ∅ | ∅
12. Noy, Natalya F.; Deborah L | 2001 | "Ontology Development 101: A Guide to Creating Your First Ontology" | ∅ | ∅ | ∅ | McGuinness | ∅ | ∅ | ∅ | ∅ | Stanford Knowledge Systems Laboratory Technical Report KSL-01-05
13. Sowa, John F. | 2000 | ∅ | Knowledge Representation: Logical, Philosophical, and Computational Foundations | ∅ | ∅ | Pacific Grove: Brooks/Cole | ∅ | ∅ | ∅ | ∅ | ∅
14. Hayes, Patrick J (ed.) | 1979 | "The Naive Physics Manifesto" | Expert Systems in the Micro-Electronic Age | ∅ | ∅ | D | ∅ | ∅ | ∅ | ∅ | Michie, 242 270; Edinburgh: Edinburgh University Press
15. Gruber, Thomas R | 1993 | "A Translation Approach to Portable Ontology Specifications" | Knowledge Acquisition | ∅ | 5.2::199–220 | ∅ | ∅ | ∅ | ∅ | ∅ | ∅
16. Bollacker, Kurt, et al. : 1247 1250 | 2008 | "Freebase: A Collaboratively Created Graph Database for Structuring Human Knowledge" | Proceedings of the 2008 ACM SIGMOD | ∅ | ∅ | ∅ | ∅ | ∅ | ∅ | ∅ | ∅
17. Ji, Shaoxiong, et al | 2022 | "A Survey on Knowledge Graphs: Representation, Acquisition, and Applications" | IEEE Transactions on Neural Networks and Learning Systems | ∅ | 33.2::494–514 | ∅ | ∅ | ∅ | ∅ | ∅ | ∅
18. Fensel, Dieter, et al | 2020 | ∅ | Knowledge Graphs: Methodology, Tools and Selected Use Cases | ∅ | ∅ | Cham: Springer | ∅ | ∅ | ∅ | ∅ | ∅
19. Studer, Rudi, V | 1998 | "Knowledge Engineering: Principles and Methods" | Data & Knowledge Engineering | ∅ | 2::161–197 | Richard Benjamins, and Dieter Fensel | ∅ | ∅ | ∅ | ∅ | 25.1
20. Lenat, Douglas B | 1995 | "CYC: A Large-Scale Investment in Knowledge Infrastructure" | Communications of the ACM | ∅ | 38.11::33–38 | ∅ | ∅ | ∅ | ∅ | ∅ | ∅
21. Auer, Sören, et al | 2007 | "DBpedia: A Nucleus for a Web of Open Data" | The Semantic Web | ∅ | ∅ | In , LNCS 4825, 722 735 | ∅ | ∅ | ∅ | ∅ | Berlin: Springer

CROSS-REFERENCE INDEX

Generated from V4 expansion plan. Last Updated: March 11, 2026

Corrections

• The Psychology of Computer Vision — invalid ISBN 0070710481 removed. No verified replacement could be found, and supplying an unverified number would be worse than none. The entry's author, title, publisher and year are unchanged.